Light Scans

In some circumstances, the database server can bypass the overhead of the buffer pool when it performs a sequential scan. Such a sequential scan is termed a light scan.

Performance advantages of using light scans instead of the buffer pool for sequential scans are as follows:

• Transfers larger blocks of data in one I/O operation

  These larger I/O blocks are usually 64 kilobytes or 128 kilobytes. To determine the I/O block size that your platform supports, see your machine notes file.

• Bypasses the overhead of the buffer pool when many pages are read
• Prevents frequently accessed pages from being forced out of the buffer pool when many sequential pages are read for a single DSS query

Light scans can be used only for sequential scans of large data tables and are the fastest means for performing these scans. System catalog tables, tables smaller than the size of the buffer pool, and tables containing varchar data do not use light scans.

Light scans occur under the following conditions:

• The optimizer chooses a sequential scan of the table.
• The number of pages in the table is greater than the number of buffers in the buffer pool.
• The isolation level obtains no lock or a shared lock on the table:
  • Dirty Read (including nonlogging databases) isolation level
  • Repeatable Read isolation level if the table has a shared or exclusive lock
  • Committed Read isolation if the table has a shared lock

Light scans do not occur under Cursor Stability isolation.

The onstat -g lsc output indicates when light scans occur, as the following sample output shows. The onstat -g lsc command displays only currently active light scans.

Light Scan Info 
descriptor  address  next_lpage  next_ppage  ppage_left  bufcnt look_aside 
6           aaa7870 3f4         200429      1488       1      N